Learning latent activities from social signals with hierarchical dirichlet processes

Understanding human activities is an important research topic, most noticeably in assisted-living and healthcare monitoring environments. Beyond simple forms of activity (e.g., an RFID event of entering a building), learning latent activities that are more semantically interpretable, such as sitting at a desk, meeting with people, or gathering with friends, remains a challenging problem. Supervised learning has been the typical modeling choice in the past. However, this requires labeled training data, is unable to predict never-seen-before activity, and fails to adapt to the continuing growth of data over time. In this chapter, we explore the use of a Bayesian nonparametric method, in particular the hierarchical Dirichlet process, to infer latent activities from sensor data acquired in a pervasive setting. Our framework is unsupervised, requires no labeled data, and is able to discover new activities as data grows. We present experiments on extracting movement and interaction activities from sociometric badge signals and show how to use them for detecting of subcommunities. Using the popular Reality Mining dataset, we further demonstrate the extraction of colocation activities and use them to automatically infer the structure of social subgroups. © 2014 Elsevier Inc. All rights reserved.

Intervention-driven predictive framework for modeling healthcare data

Assessing prognostic risk is crucial to clinical care, and critically dependent on both diagnosis and medical interventions. Current methods use this augmented information to build a single prediction rule. But this may not be expressive enough to capture differential effects of interventions on prognosis. To this end, we propose a supervised, Bayesian nonparametric framework that simultaneously discovers the latent intervention groups and builds a separate prediction rule for each intervention group. The prediction rule is learnt using diagnosis data through a Bayesian logistic regression. For inference, we develop an efficient collapsed Gibbs sampler. We demonstrate that our method outperforms baselines in predicting 30-day hospital readmission using two patient cohorts - Acute Myocardial Infarction and Pneumonia. The significance of this model is that it can be applied widely across a broad range of medical prognosis tasks. © 2014 Springer International Publishing.

Speed up health research through topic modeling of coded clinical data

Although random control trial is the gold standard in medical research, researchers are increasingly looking to alternative data sources for hypothesis generation and early-stage evidence collection. Coded clinical data are collected routinely in most hospitals. While they contain rich information directly related to the real clinical setting, they are both noisy and semantically diverse, making them difficult to analyze with conventional statistical tools. This paper presents a novel application of Bayesian nonparametric modeling to uncover latent information in coded clinical data. For a patient cohort, a Bayesian nonparametric model is used to reveal the common comorbidity groups shared by the patients and the proportion that each comorbidity group is reflected individual patient. To demonstrate the method, we present a case study based on hospitalization coding from an Australian hospital. The model recovered 15 comorbidity groups among 1012 patients hospitalized during a month. When patients from two areas of unequal socio-economic status were compared, it reveals higher prevalence of diverticular disease in the region of lower socio-economic status. The study builds a convincing case for routine coded data to speed up hypothesis generation.

Continuous discovery of co-location contexts from Bluetooth data

The discovery of contexts is important for context-aware applications in pervasive computing. This is a challenging problem because of the stream nature of data, the complexity and changing nature of contexts. We propose a Bayesian nonparametric model for the detection of co-location contexts from Bluetooth signals. By using an Indian buffet process as the prior distribution, the model can discover the number of contexts automatically. We introduce a novel fixed-lag particle filter that processes data incrementally. This sampling scheme is especially suitable for pervasive computing as the computational requirements remain constant in spite of growing data. We examine our model on a synthetic dataset and two real world datasets. To verify the discovered contexts, we compare them to the communities detected by the Louvain method, showing a strong correlation between the results of the two methods. Fixed-lag particle filter is compared with Gibbs sampling in terms of the normalized factorization error that shows a close performance between the two inference methods. As fixed-lag particle filter processes a small chunk of data when it comes and does not need to be restarted, its execution time is significantly shorter than that of Gibbs sampling.

Streaming variational inference for dirichlet process mixtures

Bayesian nonparametric models are theoretically suitable to learn streaming data due to their complexity relaxation to the volume of observed data. However, most of the existing variational inference algorithms are not applicable to streaming applications since they re-quire truncation on variational distributions. In this paper, we present two truncation-free variational algorithms, one for mix-membership inference called TFVB (truncation-free variational Bayes), and the other for hard clustering inference called TFME (truncation-free maximization expectation). With these algorithms, we further developed a streaming learning framework for the popular Dirichlet process mixture (DPM) models. Our ex-periments demonstrate the usefulness of our framework in both synthetic and real-world data.

Data clustering using side information dependent Chinese restaurant processes

Side information, or auxiliary information associated with documents or image content, provides hints for clustering. We propose a new model, side information dependent Chinese restaurant process, which exploits side information in a Bayesian nonparametric model to improve data clustering. We introduce side information into the framework of distance dependent Chinese restaurant process using a robust decay function to handle noisy side information. The threshold parameter of the decay function is updated automatically in the Gibbs sampling process. A fast inference algorithm is proposed. We evaluate our approach on four datasets: Cora, 20 Newsgroups, NUS-WIDE and one medical dataset. Types of side information explored in this paper include citations, authors, tags, keywords and auxiliary clinical information. The comparison with the state-of-the-art approaches based on standard performance measures (NMI, F1) clearly shows the superiority of our approach.

Keeping up with innovation: a predictive framework for modeling healthcare data with evolving clinical interventions

Medical outcomes are inexorably linked to patient illness and clinical interventions. Interventions change the course of disease, crucially determining outcome. Traditional outcome prediction models build a single classifier by augmenting interventions with disease information. Interventions, however, differentially affect prognosis, thus a single prediction rule may not suffice to capture variations. Interventions also evolve over time as more advanced interventions replace older ones. To this end, we propose a Bayesian nonparametric, supervised framework that models a set of intervention groups through a mixture distribution building a separate prediction rule for each group, and allows the mixture distribution to change with time. This is achieved by using a hierarchical Dirichlet process mixture model over the interventions. The outcome is then modeled as conditional on both the latent grouping and the disease information through a Bayesian logistic regression. Experiments on synthetic and medical cohorts for 30-day readmission prediction demonstrate the superiority of the proposed model over clinical and data mining baselines.

SECC: simultaneous extraction of context and community from pervasive signals

Understanding user contexts and group structures plays a central role in pervasive computing. These contexts and community structures are complex to mine from data collected in the wild due to the unprecedented growth of data, noise, uncertainties and complexities. Typical existing approaches would first extract the latent patterns to explain the human dynamics or behaviors and then use them as the way to consistently formulate numerical representations for community detection, often via a clustering method. While being able to capture high-order and complex representations, these two steps are performed separately. More importantly, they face a fundamental difficulty in determining the correct number of latent patterns and communities. This paper presents an approach that seamlessly addresses these challenges to simultaneously discover latent patterns and communities in a unified Bayesian nonparametric framework. Our Simultaneous Extraction of Context and Community (SECC) model roots in the nested Dirichlet process theory which allows nested structure to be built to explain data at multiple levels. We demonstrate our framework on three public datasets where the advantages of the proposed approach are validated.

Inferência bayesiana não-paramétrica para elicitação da função de contabilidade

Pós-graduação em Matematica Aplicada e Computacional - FCT

Alarm systems and catastrophes from a diverse point of view

Using a chain of urns, we build a Bayesian nonparametric alarm system to predict catastrophic events, such as epidemics, black outs, etc. Differently from other alarm systems in the literature, our model is constantly updated on the basis of the available information, according to the Bayesian paradigm. The papers contains both theoretical and empirical results. In particular, we test our alarm system on a well-known time series of sunspots.

Learning multi-faceted activities from heterogeneous data with the product space hierarchical Dirichlet processes

Hierarchical Dirichlet processes (HDP) was originally designed and experimented for a single data channel. In this paper we enhanced its ability to model heterogeneous data using a richer structure for the base measure being a product-space. The enhanced model, called Product Space HDP (PS-HDP), can (1) simultaneously model heterogeneous data from multiple sources in a Bayesian nonparametric framework and (2) discover multilevel latent structures from data to result in different types of topics/latent structures that can be explained jointly. We experimented with the MDC dataset, a large and real-world data collected from mobile phones. Our goal was to discover identity–location– time (a.k.a who-where-when) patterns at different levels (globally for all groups and locally for each group). We provided analysis on the activities and patterns learned from our model, visualized, compared and contrasted with the ground-truth to demonstrate the merit of the proposed framework. We further quantitatively evaluated and reported its performance using standard metrics including F1-score, NMI, RI, and purity. We also compared the performance of the PS-HDP model with those of popular existing clustering methods (including K-Means, NNMF, GMM, DP-Means, and AP). Lastly, we demonstrate the ability of the model in learning activities with missing data, a common problem encountered in pervasive and ubiquitous computing applications.

Simple approximate MAP inference for Dirichlet processes mixtures

The Dirichlet process mixture model (DPMM) is a ubiquitous, flexible Bayesian nonparametric statistical model. However, full probabilistic inference in this model is analytically intractable, so that computationally intensive techniques such as Gibbs sampling are required. As a result, DPMM-based methods, which have considerable potential, are restricted to applications in which computational resources and time for inference is plentiful. For example, they would not be practical for digital signal processing on embedded hardware, where computational resources are at a serious premium. Here, we develop a simplified yet statistically rigorous approximate maximum a-posteriori (MAP) inference algorithm for DPMMs. This algorithm is as simple as DP-means clustering, solves the MAP problem as well as Gibbs sampling, while requiring only a fraction of the computational effort. (For freely available code that implements the MAP-DP algorithm for Gaussian mixtures see http://www.maxlittle.net/.) Unlike related small variance asymptotics (SVA), our method is non-degenerate and so inherits the “rich get richer” property of the Dirichlet process. It also retains a non-degenerate closed-form likelihood which enables out-of-sample calculations and the use of standard tools such as cross-validation. We illustrate the benefits of our algorithm on a range of examples and contrast it to variational, SVA and sampling approaches from both a computational complexity perspective as well as in terms of clustering performance. We demonstrate the wide applicabiity of our approach by presenting an approximate MAP inference method for the infinite hidden Markov model whose performance contrasts favorably with a recently proposed hybrid SVA approach. Similarly, we show how our algorithm can applied to a semiparametric mixed-effects regression model where the random effects distribution is modelled using an infinite mixture model, as used in longitudinal progression modelling in population health science. Finally, we propose directions for future research on approximate MAP inference in Bayesian nonparametrics.

Dirichlet process mixture models with pairwise constraints for data clustering

The Dirichlet process mixture (DPM) model, a typical Bayesian nonparametric model, can infer the number of clusters automatically, and thus performing priority in data clustering. This paper investigates the influence of pairwise constraints in the DPM model. The pairwise constraint, known as two types: must-link (ML) and cannot-link (CL) constraints, indicates the relationship between two data points. We have proposed two relevant models which incorporate pairwise constraints: the constrained DPM (C-DPM) and the constrained DPM with selected constraints (SC-DPM). In C-DPM, the concept of chunklet is introduced. ML constraints are compiled into chunklets and CL constraints exist between chunklets. We derive the Gibbs sampling of the C-DPM based on chunklets. We further propose a principled approach to select the most useful constraints, which will be incorporated into the SC-DPM. We evaluate the proposed models based on three real datasets: 20 Newsgroups dataset, NUS-WIDE image dataset and Facebook comments datasets we collected by ourselves. Our SC-DPM performs priority in data clustering. In addition, our SC-DPM can be potentially used for short-text clustering.

A simultaneous extraction of context and community from pervasive signals using nested dirichlet process

Understanding user contexts and group structures plays a central role in pervasive computing. These contexts and community structures are complex to mine from data collected in the wild due to the unprecedented growth of data, noise, uncertainties and complexities. Typical existing approaches would first extract the latent patterns to explain human dynamics or behaviors and then use them as a way to consistently formulate numerical representations for community detection, often via a clustering method. While being able to capture high-order and complex representations, these two steps are performed separately. More importantly, they face a fundamental difficulty in determining the correct number of latent patterns and communities. This paper presents an approach that seamlessly addresses these challenges to simultaneously discover latent patterns and communities in a unified Bayesian nonparametric framework. Our Simultaneous Extraction of Context and Community (SECC) model roots in the nested Dirichlet process theory which allows a nested structure to be built to summarize data at multiple levels. We demonstrate our framework on five datasets where the advantages of the proposed approach are validated.